Genetic algorithm for scheduling of IRIS (Increased reward with increased service) tasks

In this paper we consider on-line reward-based scheduling for real-time tasks having hard deadlines on a single processor, which arrive in arbitrary manner. The scheme used is Increased Reward for Increased service (IRIS) where the reward of the task is an increasing function of the amount of service it receives. Our aim is to maximize average reward function. A genetic algorithm is proposed to find the optimal schedule of tasks such that it maximizes the net reward function. A single algorithm finds out the accrued execution time of each task and schedules it.